CLIP ViT-B/32 for Aether-Kit (Forge/clip export)

.aether/aether.json

@@ -0,0 +1,220 @@
+{
+  "schema": 1,
+  "id": "clip-vit-b32",
+  "task": "image-text-embedding",
+  "driver": "clip-joint@1",
+  "variant": {
+    "platform": "macos"
+  },
+  "assets": {
+    "main": {
+      "subtree": "",
+      "files": [
+        {
+          "path": "LICENSE",
+          "bytes": 1064,
+          "sha256": "987e63b32f6c89ff5160e429458a872ff048e6860b590a3912e938f9da8f14db"
+        },
+        {
+          "path": "NOTICE",
+          "bytes": 1129,
+          "sha256": "65fb3e46dcdcc7b630ef96ece0da2fb177e7a5d348c57091fe345c0d2d5def18"
+        },
+        {
+          "path": "README.md",
+          "bytes": 1300,
+          "sha256": "ae045dc8fc978ef9a972b3442286128da3cc00f67e997254c8e77824cde83fca"
+        },
+        {
+          "path": "clip_vit_b32_image.aimodel/main.hash",
+          "bytes": 32,
+          "sha256": "d1c0b0f0323349a4bd9c48460bc7681a560c8ebb14bc00aa800f165a645f9f79"
+        },
+        {
+          "path": "clip_vit_b32_image.aimodel/main.mlirb",
+          "bytes": 175768383,
+          "sha256": "d1e96135d26898b55cad462a2da4e75ea41149b4f79ca608cb01a648d24ea4e8"
+        },
+        {
+          "path": "clip_vit_b32_image.aimodel/metadata.json",
+          "bytes": 350,
+          "sha256": "be18860a32f3c74e9faec571fe89a8cc705b123ee5b52a3f4a43d8e2a878459d"
+        },
+        {
+          "path": "clip_vit_b32_text.aimodel/main.hash",
+          "bytes": 32,
+          "sha256": "5107534982347924da3ed4aded6564ac8c17be238e8b37021520bc2a650ed8f9"
+        },
+        {
+          "path": "clip_vit_b32_text.aimodel/main.mlirb",
+          "bytes": 127030587,
+          "sha256": "69a0127b96cac208d3ef71fdf0b6e832997621008abd2ab64cc1d8b87b0d4e01"
+        },
+        {
+          "path": "clip_vit_b32_text.aimodel/metadata.json",
+          "bytes": 367,
+          "sha256": "6a4f126ad607f3aea64335f49a3f1cd8f126d9c6c0775ecd2c77691c4c2e7511"
+        },
+        {
+          "path": "tokenizer/merges.txt",
+          "bytes": 524657,
+          "sha256": "f526393189112391ce6f9795d4695f704121ce452c3aad1f5335cc41337eba85"
+        },
+        {
+          "path": "tokenizer/special_tokens_map.json",
+          "bytes": 389,
+          "sha256": "f8c0d6c39aee3f8431078ef6646567b0aba7f2246e9c54b8b99d55c22b707cbf"
+        },
+        {
+          "path": "tokenizer/tokenizer.json",
+          "bytes": 2224041,
+          "sha256": "b556ac8c99757ffb677208af34bc8c6721572114111a6e0aaf5fa69ff0b8d842"
+        },
+        {
+          "path": "tokenizer/tokenizer_config.json",
+          "bytes": 592,
+          "sha256": "34b7336e4bee12e0a9730eaf5189f582ef3c3eea5027f65730e5717256755aad"
+        },
+        {
+          "path": "tokenizer/vocab.json",
+          "bytes": 862328,
+          "sha256": "5047b556ce86ccaf6aa22b3ffccfc52d391ea4accdab9c2f2407da5b742d4363"
+        }
+      ]
+    }
+  },
+  "graphs": {
+    "imageEncoder": {
+      "file": "main:clip_vit_b32_image.aimodel",
+      "function": "main",
+      "inputs": {
+        "image": "image"
+      },
+      "outputs": {
+        "imageEmbedding": "image_embedding"
+      },
+      "compute": "gpu",
+      "expectFrequentReshapes": false
+    },
+    "textEncoder": {
+      "file": "main:clip_vit_b32_text.aimodel",
+      "function": "batch1",
+      "buckets": {
+        "axis": "batch",
+        "rule": "smallestFitting",
+        "entries": [
+          {
+            "value": 1,
+            "function": "batch1"
+          },
+          {
+            "value": 8,
+            "function": "batch8"
+          }
+        ]
+      },
+      "inputs": {
+        "tokens": "tokens"
+      },
+      "outputs": {
+        "textEmbedding": "text_embedding"
+      },
+      "compute": "gpu",
+      "expectFrequentReshapes": false,
+      "limits": {
+        "staticSequence": 77
+      }
+    }
+  },
+  "tokenizer": {
+    "source": {
+      "dir": "main:tokenizer"
+    },
+    "chatTemplate": null,
+    "renderer": "clip-bpe",
+    "addSpecialTokens": true,
+    "text": {
+      "padId": 49407,
+      "padSide": "right",
+      "padTo": 77,
+      "truncation": "right",
+      "maskRule": "none",
+      "unusedBatchRows": [
+        49406,
+        49407
+      ]
+    },
+    "x-notes": "CLIP BPE (NFC, whitespace collapsed, lowercase; `</w>` end-of-word): BOS 49406, at most 75 pieces, EOS 49407, right-padded with 49407 to 77, as transformers' CLIPTokenizer does with padding='max_length', truncation=True. Only <|startoftext|> and <|endoftext|> are added tokens: '!' is an ordinary piece here ('hello!' -> 3306, 256), unlike SD 2.x's tokenizer, where '!' is the pad token 0. The graph takes no attention mask (causal attention makes it redundant) and pools the first EOS, so any pad id works. Rows of a bucket beyond the texts are filled with [BOS, EOS, then pad] and their outputs discarded."
+  },
+  "preprocess": {
+    "image": {
+      "resize": {
+        "mode": "centerCrop",
+        "size": [
+          224,
+          224
+        ],
+        "filter": "pillowBicubic",
+        "shortestSide": 224
+      },
+      "colorSpace": "sRGB",
+      "channelOrder": "RGB",
+      "scale": 0.00392156862745098,
+      "mean": [
+        0.0,
+        0.0,
+        0.0
+      ],
+      "std": [
+        1.0,
+        1.0,
+        1.0
+      ],
+      "layout": "NCHW",
+      "hostDType": "float32",
+      "orientation": "exif",
+      "x-notes": "Upstream (openai/CLIP `_transform`: torchvision Resize(224, BICUBIC) + CenterCrop(224) on a PIL image): resize the shortest side to 224 with Pillow BICUBIC (long side int(224 * long / short)), center-crop 224 x 224 (offsets round((side - 224) / 2)), /255. transformers 5's CLIPImageProcessor floors the crop offset and resizes with torch, so it can differ by a pixel of offset on odd margins. OpenAI CLIP normalization (mean 0.48145466/0.4578275/0.40821073, std 0.26862954/0.26130258/0.27577711) is in the graph: feed RGB on [0,1]. The graph input is float16; the host builds float32 and converts."
+    }
+  },
+  "postprocess": {
+    "kind": "embedding",
+    "embedding": {
+      "normalized": "inGraph",
+      "pooling": "inGraph",
+      "similarity": "cosine",
+      "dimension": 512,
+      "logitScale": 100.00000762939453
+    },
+    "x-notes": "imageEmbedding [1,512] and textEmbedding [B,512] are unit-length float32 (L2-normalized in the graph), so cosine similarity is their dot product. Zero-shot classification: softmax over labels of logitScale * cosine(image, text(label)); logitScale is upstream's logit_scale.exp(). OpenAI's prompt for ImageNet-style labels is 'a photo of a {label}.'."
+  },
+  "precision": {
+    "weights": "fp16",
+    "activations": "fp16",
+    "components": {
+      "towers": "fp16 (both transformers and their projections)",
+      "normalization": "fp32 (input mean/std and the output L2 normalization)",
+      "io": "image float16, tokens int32; embeddings float32"
+    }
+  },
+  "license": "MIT",
+  "x-license": {
+    "files": [
+      "LICENSE",
+      "NOTICE"
+    ],
+    "modified": true,
+    "x-notes": "OpenAI's model card scopes CLIP to research use: it calls any deployed use currently out of scope without task-specific testing, and surveillance and facial recognition always out of scope. That is guidance, not a license term."
+  },
+  "provenance": {
+    "source": "openai/clip-vit-base-patch32",
+    "sourceRevision": "3d74acf9a28c67741b2f4f2ea7635f0aaf6f0268",
+    "producer": "aether-kit Forge/clip/export_clip.py",
+    "exportScript": "d6b32fd5d8a8ccda2e608cc6b73446be620904aa9c9307ed5e4f6ffbc38155b0",
+    "refs": {
+      "upstreamCode": "transformers 5.22.1 CLIPModel",
+      "license": "openai/CLIP@d05afc436d78f1c48dc0dbf8e5980a9d471f35f6:LICENSE",
+      "forgeLockSHA256": "6729c123d0399129ae8ed3db306a9a0b680da0f9b61890218c328a7ac6e1920a",
+      "exportReport": "export-report.json"
+    }
+  }
+}

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+clip_vit_b32_image.aimodel/main.mlirb filter=lfs diff=lfs merge=lfs -text
+clip_vit_b32_text.aimodel/main.mlirb filter=lfs diff=lfs merge=lfs -text

LICENSE

@@ -0,0 +1,22 @@
+MIT License
+
+Copyright (c) 2021 OpenAI
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+

NOTICE

@@ -0,0 +1,15 @@
+CLIP ViT-B/32 (Aether-Kit Core AI export)
+
+This bundle is a modified version of openai/clip-vit-base-patch32 at revision 3d74acf9a28c67741b2f4f2ea7635f0aaf6f0268,
+by OpenAI (CLIP; https://github.com/openai/CLIP), licensed under the MIT License (LICENSE, from
+openai/CLIP at commit d05afc436d78f1c48dc0dbf8e5980a9d471f35f6). The tokenizer files in tokenizer/ are the checkpoint's, unchanged.
+
+Modifications: Aether-Kit's Forge exporter re-expressed the model, as implemented by Hugging Face transformers'
+CLIPModel (Apache-2.0), as two static Core AI graphs: the image tower (224 x 224 input, CLIP normalization
+folded in) and the text tower (77 tokens, static batches 1 and 8, pooled at the first end-of-text token
+without an attention mask), each with its projection and an in-graph L2 normalization, weights in float16.
+It is not an official release of CLIP.
+
+OpenAI's model card (https://github.com/openai/CLIP/blob/main/model-card.md) describes CLIP as a research
+model: it considers any deployed use currently out of scope without thorough task-specific testing, and
+surveillance and facial recognition always out of scope.

README.md

@@ -0,0 +1,22 @@
+---
+license: mit
+base_model: openai/clip-vit-base-patch32
+pipeline_tag: zero-shot-image-classification
+tags: [aether-kit, core-ai, clip, image-text-embedding, zero-shot-image-classification]
+---
+
+# CLIP ViT-B/32 for Aether-Kit (Core AI)
+
+[openai/clip-vit-base-patch32](https://huggingface.co/openai/clip-vit-base-patch32) at revision `3d74acf9`, exported by
+[Aether-Kit](https://github.com/bubstackdev/aether-kit)'s Forge exporter (`Forge/clip/`) as two static Core
+AI graphs: the image tower (an RGB image resized to 224 on its shortest side and center-cropped, on [0, 1];
+CLIP normalization is in the graph) and the text tower (77 CLIP BPE tokens, batches of 1 or 8). Each returns
+unit-length 512-wide embeddings in float32; the towers run in float16. Cosine similarity is the dot product, and
+zero-shot classification is a softmax over 100 x cosine. Load it with Aether-Kit's `AetherVision`
+(`clip-joint@1`, macOS); the manifest is `.aether/aether.json`. Gated against the
+upstream implementation on CPU in float32 (see `export-report.json`).
+
+OpenAI's [model card](https://github.com/openai/CLIP/blob/main/model-card.md) describes CLIP as a research
+model and considers untested deployed use, surveillance and facial recognition out of scope.
+
+MIT (`LICENSE`). A modified version (see `NOTICE`).

SHA256SUMS

@@ -0,0 +1,16 @@
+ab851888c05356d89b9eb7ef1bfe1b017e282b02055926847ef76d360510b035  .aether/aether.json
+987e63b32f6c89ff5160e429458a872ff048e6860b590a3912e938f9da8f14db  LICENSE
+65fb3e46dcdcc7b630ef96ece0da2fb177e7a5d348c57091fe345c0d2d5def18  NOTICE
+ae045dc8fc978ef9a972b3442286128da3cc00f67e997254c8e77824cde83fca  README.md
+d1c0b0f0323349a4bd9c48460bc7681a560c8ebb14bc00aa800f165a645f9f79  clip_vit_b32_image.aimodel/main.hash
+d1e96135d26898b55cad462a2da4e75ea41149b4f79ca608cb01a648d24ea4e8  clip_vit_b32_image.aimodel/main.mlirb
+be18860a32f3c74e9faec571fe89a8cc705b123ee5b52a3f4a43d8e2a878459d  clip_vit_b32_image.aimodel/metadata.json
+5107534982347924da3ed4aded6564ac8c17be238e8b37021520bc2a650ed8f9  clip_vit_b32_text.aimodel/main.hash
+69a0127b96cac208d3ef71fdf0b6e832997621008abd2ab64cc1d8b87b0d4e01  clip_vit_b32_text.aimodel/main.mlirb
+6a4f126ad607f3aea64335f49a3f1cd8f126d9c6c0775ecd2c77691c4c2e7511  clip_vit_b32_text.aimodel/metadata.json
+1730e2280967bf6099f2ca4f1886362d1dec6135b74dfe08aaf55ee9ccdb1dbd  export-report.json
+f526393189112391ce6f9795d4695f704121ce452c3aad1f5335cc41337eba85  tokenizer/merges.txt
+f8c0d6c39aee3f8431078ef6646567b0aba7f2246e9c54b8b99d55c22b707cbf  tokenizer/special_tokens_map.json
+b556ac8c99757ffb677208af34bc8c6721572114111a6e0aaf5fa69ff0b8d842  tokenizer/tokenizer.json
+34b7336e4bee12e0a9730eaf5189f582ef3c3eea5027f65730e5717256755aad  tokenizer/tokenizer_config.json
+5047b556ce86ccaf6aa22b3ffccfc52d391ea4accdab9c2f2407da5b742d4363  tokenizer/vocab.json

clip_vit_b32_image.aimodel/main.hash

@@ -0,0 +1 @@
+��a5�h��\�F*-��^�I�����
�H�N��

clip_vit_b32_image.aimodel/main.mlirb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1e96135d26898b55cad462a2da4e75ea41149b4f79ca608cb01a648d24ea4e8
+size 175768383

clip_vit_b32_image.aimodel/metadata.json

@@ -0,0 +1,8 @@
+{
+  "assetVersion" : "2.0",
+  "author" : "OpenAI (CLIP); Core AI export: Aether-Kit Forge",
+  "description" : "CLIP ViT-B\/32 image tower (openai\/clip-vit-base-patch32@3d74acf9a28c): image [1,3,224,224] RGB on [0,1] -> unit image_embedding [1,512].",
+  "producer" : "coreai-core 1.0.0b2",
+  "license" : "MIT",
+  "creationDate" : "20261106T195539Z"
+}

clip_vit_b32_text.aimodel/main.mlirb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69a0127b96cac208d3ef71fdf0b6e832997621008abd2ab64cc1d8b87b0d4e01
+size 127030587

clip_vit_b32_text.aimodel/metadata.json

@@ -0,0 +1,8 @@
+{
+  "producer" : "coreai-core 1.0.0b2",
+  "assetVersion" : "2.0",
+  "description" : "CLIP ViT-B\/32 text tower (openai\/clip-vit-base-patch32@3d74acf9a28c): tokens [B,77] int32 (B = 1 or 8, one function each) -> unit text_embedding [B,512].",
+  "creationDate" : "20261106T195541Z",
+  "license" : "MIT",
+  "author" : "OpenAI (CLIP); Core AI export: Aether-Kit Forge"
+}

export-report.json

@@ -0,0 +1,129 @@
+{
+  "schema": 1,
+  "id": "clip-vit-b32",
+  "command": [
+    "Forge/clip/export_clip.py",
+    "--output",
+    "/Users/zakkeown/Code/bubstackdev/aether/forge-outputs/clip-vit-b32/bundle"
+  ],
+  "exportedAt": "2026-11-06T19:55:37.010068+00:00",
+  "host": {
+    "os": "26B123",
+    "machine": "arm64",
+    "python": "3.12.16"
+  },
+  "versions": {
+    "torch": "2.11.0",
+    "torchvision": "0.26.0",
+    "transformers": "5.22.1",
+    "coreai-torch": "0.4.2",
+    "coreai-core": "1.0.0b2",
+    "numpy": "2.5.4"
+  },
+  "forgeLockSHA256": "6729c123d0399129ae8ed3db306a9a0b680da0f9b61890218c328a7ac6e1920a",
+  "forgePyprojectSHA256": "652f31b5ec9485de2ac31e1bba00433f1ab16127f3b56822eeca5368e4f2bc77",
+  "scriptSHA256": {
+    "bench_clip.py": "67dfea418af25f5faa015507af6176ab6b13c869efb70ef92700a2e165619adc",
+    "check_clip.py": "1d2286d8c9ff084f54675edca517fbe24362ba647075dc3d35fd1b6f32918471",
+    "clip_graphs.py": "d715a141ac2befbe243e0c49b5e5465ed748fef505164de8f125fc6109dbd8c3",
+    "export_clip.py": "d6b32fd5d8a8ccda2e608cc6b73446be620904aa9c9307ed5e4f6ffbc38155b0",
+    "fetch_upstream.py": "39eae947367748d35762783e4c1ccaf837e782fc5a5fe38c6c70a3ada05e3d2f",
+    "reference_clip.py": "ec9ca1690b952804aa365058f1fbce82fd64d77f4ea2b02971ddb01e3a0c057f",
+    "selftest_clip.py": "0952e4d45ad836c2c6b0ac1b04b93d2e1fd03dd58f23cbf5c77a91f933b1fad5"
+  },
+  "source": {
+    "repository": "openai/clip-vit-base-patch32",
+    "revision": "3d74acf9a28c67741b2f4f2ea7635f0aaf6f0268",
+    "snapshot": "/Users/zakkeown/.cache/huggingface/hub/models--openai--clip-vit-base-patch32/snapshots/3d74acf9a28c67741b2f4f2ea7635f0aaf6f0268",
+    "license": {
+      "repository": "openai/CLIP",
+      "commit": "d05afc436d78f1c48dc0dbf8e5980a9d471f35f6"
+    },
+    "verified": {
+      "README.md": "8f0e8c678e0afe5f384f3bf8a6694c1cb53c41e52f86277967b6c5354cb0f9f4",
+      "config.json": "b575ef3c36f2a057fa19e221650105052d61cc9c1a972ec15019c6261ec98770",
+      "merges.txt": "f526393189112391ce6f9795d4695f704121ce452c3aad1f5335cc41337eba85",
+      "preprocessor_config.json": "910e70b3956ac9879ebc90b22fb3bc8a75b6a0677814500101a4c072bd7857bd",
+      "pytorch_model.bin": "a63082132ba4f97a80bea76823f544493bffa8082296d62d71581a4feff1576f",
+      "special_tokens_map.json": "f8c0d6c39aee3f8431078ef6646567b0aba7f2246e9c54b8b99d55c22b707cbf",
+      "tokenizer.json": "b556ac8c99757ffb677208af34bc8c6721572114111a6e0aaf5fa69ff0b8d842",
+      "tokenizer_config.json": "34b7336e4bee12e0a9730eaf5189f582ef3c3eea5027f65730e5717256755aad",
+      "vocab.json": "5047b556ce86ccaf6aa22b3ffccfc52d391ea4accdab9c2f2407da5b742d4363",
+      "upstream:LICENSE": "987e63b32f6c89ff5160e429458a872ff048e6860b590a3912e938f9da8f14db",
+      "upstream:model-card.md": "7baf04f60c6234b301ec2c9ca39e67a3ca54b47c05e9509bddf732cbcbec8b7f",
+      "images:skimage/data/astronaut.png": "88431cd9653ccd539741b555fb0a46b61558b301d4110412b5bc28b5e3ea6cb5",
+      "images:skimage/data/camera.png": "b0793d2adda0fa6ae899c03989482bff9a42d3d5690fc7e3648f2795d730c23a",
+      "images:skimage/data/chelsea.png": "596aa1e7cb875eb79f437e310381d26b338a81c2da23439704a73c4651e8c4bb",
+      "images:skimage/data/coffee.png": "cc02f8ca188b167c775a7101b5d767d1e71792cf762c33d6fa15a4599b5a8de7",
+      "images:skimage/data/motorcycle_right.png": "5fc913ae870e42a4b662314bc904d1786bcad8e2f0b9b67dba5a229406357797",
+      "images:skimage/data/rocket.jpg": "c2dd0de7c538df8d111e479619b129464d0269d0ae5fd18ca91d33a7fdfea95c"
+    }
+  },
+  "precisionName": "float16",
+  "precision": "towers float16; normalization float32; image float16 and tokens int32 in, embeddings float32 out; no quantization",
+  "inputShapes": {
+    "image": [
+      1,
+      3,
+      224,
+      224
+    ],
+    "tokens": {
+      "batch1": [
+        1,
+        77
+      ],
+      "batch8": [
+        8,
+        77
+      ]
+    }
+  },
+  "compute": "gpu",
+  "conversion": {
+    "imageEncoder": {
+      "exportSeconds": 0.8,
+      "convertSeconds": 0.3,
+      "bytes": 175768765,
+      "inputs": [
+        "image"
+      ],
+      "outputs": [
+        "image_embedding"
+      ],
+      "functions": [
+        "main"
+      ]
+    },
+    "textEncoder": {
+      "exportSeconds": 1.3,
+      "convertSeconds": 0.4,
+      "bytes": 127030986,
+      "inputs": [
+        "tokens"
+      ],
+      "outputs": [
+        "text_embedding"
+      ],
+      "functions": [
+        "batch1",
+        "batch8"
+      ]
+    }
+  },
+  "stages": {
+    "verify-sources": 1.64,
+    "export-image": 1.21,
+    "export-text": 1.89,
+    "package": 0.11
+  },
+  "logitScale": 100.00000762939453,
+  "peakRSSGB": 1.59,
+  "wallSeconds": 4.9,
+  "bundleBytes": 306415251,
+  "assetBytes": {
+    "imageEncoder": 175768765,
+    "textEncoder": 127030986
+  },
+  "aetherJSONSHA256": "ab851888c05356d89b9eb7ef1bfe1b017e282b02055926847ef76d360510b035"
+}

forge-log.jsonl

@@ -0,0 +1,8 @@
+{"message": "start", "stage": "verify-sources", "t": 1793994937.025}
+{"message": "done", "peakRSSGB": 0.99, "seconds": 1.64, "stage": "verify-sources", "t": 1793994938.664}
+{"message": "start", "stage": "export-image", "t": 1793994938.665}
+{"message": "done", "peakRSSGB": 1.49, "seconds": 1.21, "stage": "export-image", "t": 1793994939.869}
+{"message": "start", "stage": "export-text", "t": 1793994939.87}
+{"message": "done", "peakRSSGB": 1.59, "seconds": 1.89, "stage": "export-text", "t": 1793994941.76}
+{"message": "start", "stage": "package", "t": 1793994941.76}
+{"message": "done", "peakRSSGB": 1.59, "seconds": 0.11, "stage": "package", "t": 1793994941.87}

tokenizer/special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": {"content": "<|startoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "eos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "unk_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "pad_token": "<|endoftext|>"}

tokenizer/tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "bos_token": {"content": "<|startoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "eos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "pad_token": "<|endoftext|>", "add_prefix_space": false, "errors": "replace", "do_lower_case": true, "name_or_path": "./clip_ViT_B_32/", "model_max_length": 77}